Theorem-of-Thought improves LLM reasoning accuracy on WebOfLies and MultiArith by selecting among three differently styled reasoning chains via NLI-based coherence scoring.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Theorem-of-Thought: A Multi-Agent Framework for Abductive, Deductive, and Inductive Reasoning in Language Models
Theorem-of-Thought improves LLM reasoning accuracy on WebOfLies and MultiArith by selecting among three differently styled reasoning chains via NLI-based coherence scoring.